Anomaly detection of data and topology patterns in WSNs
| Autor: | Trevor P. Martin, George Oikonomou, Robert Zakrzewski |
|---|---|
| Rok vydání: | 2021 |
| Předmět: |
Sensor networks
Computer science Topology (electrical circuits) Anomaly detection Cyber-security Network topology Topology Graph Fault detection and isolation Tree (data structure) Machine learning Enhanced Data Rates for GSM Evolution Anomaly (physics) Data pattern Fault detection Wireless sensor network |
| Zdroj: | DCOSS Zakrzewski, R, Martin, T P & Oikonomou, G 2021, Anomaly detection of data and topology patterns in WSNs . in 17th International Conference on Distributed Computing in Sensor Systems (DCOSS) . Institute of Electrical and Electronics Engineers (IEEE), pp. 535-542, 2021 17th International Conference on Distributed Computing in Sensor Systems (DCOSS), Pafos, Cyprus, 14/07/21 . https://doi.org/10.1109/DCOSS52077.2021.00087 |
| DOI: | 10.1109/dcoss52077.2021.00087 |
| Popis: | Wireless sensor networks are often distributedwhich makes detection of cyber-attacks or misconfiguration hard.Topology and data patterns change may result from attacksleading to the compromise of data and service availability orindicate operational problems. Graphs are often used to modeltopology and data paths to describe and compare state of asystem. For anomaly detection, the definition of normal patterns,deviation from normal, and criteria when to declare anomalyare required. In this contribution the process of acquisition ofnormal patterns (ground truth), and criteria when to declareanomaly based on graph comparison are proposed. The anomalydetection is suitable for deployment at the edge of a network.Finally, the inability to define all security threats is addressedby a custom tree-based classifier which only requires normalpatterns for training. A simulated wireless sensor network wasused to acquire data and apply the method. Our experimentsshow that data and topology change can be detected at the edgeof a network. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|